Techniq es in Wate shed Management”
M odu le 6 – ( L2 2 – L2 6 ) : “ Use of M ode r n Te ch n iqu e s in W a t e r sh e d M a n a ge m e n t ” Te ch n iq e s in W a t e sh e d M a n a ge m e n t ” Applica t ion s of Ge ogr a ph ica l I n for m a t ion Syst e m a n d Re m ot e Se n sin g in W a t e r sh e d M a n a ge m e n t , Role of g g , D e cision Su ppor t Syst e m in W a t e r sh e d M a n a ge m e n t Applica t ion s of Kn ow le dge 2 6 2 6 B Ba se d M ode ls in W a t e r sh e d d M d l i W t h d
Applications of Knowledge Based L26– L26 Models in Watershed Management d l h d
Topics Cove r e d Topics Cove r e d
Knowledge based Modeling, Multi criteria Knowledge based Modeling, Multi criteria decision analysis, Fuzzy Logic based decision analysis Fuzzy Logic based decision analysis, Fuzzy Logic based decision analysis Fuzzy Logic based modeling, Fuzzy systems, Applications of modeling, Fuzzy systems, Applications of Knowledge based Systems in Watershed Knowledge based Systems in Watershed K K l d l d b b d S d S i i W W h d h d Management. Management.
Knowledge based model; Multi criteria Knowledge based model; Multi criteria
Keywords: Keywords: decision analysis; Fuzzy logic based modeling decision analysis; Fuzzy logic based modeling decision analysis; Fuzzy logic based modeling decision analysis; Fuzzy logic based modeling .. .. Knowledge Based Model (KBM) Knowledge Based Model (KBM)
Knowledge-based systems are computer systems that
are programmed to imitate the human problem are programmed to imitate the human problem- solving ability by means of artificial intelligence (AI) tools.
Human reasoning Human reasoning using natural language can be using natural language can be
reproduced in knowledge-based systems through AI tools.
In KBM , knowledge can be: knowledge derived from
basic analysis, experts’ knowledge collected through surveys, & heuristic information from the field. surveys, & heuristic information from the field.
Experts’ knowledge and heuristic information related
to the specific problems are generally stored in the form of a rule-base. form of a rule base Knowledge Based Modeling Knowledge Based Modeling
A knowledge base is an organized body of knowledge that provides a formal logical specification for the that provides a formal logical specification for the interpretation of information
In the knowledge-based modeling approach , watershed assessment is a watershed assessment is a multi-criteria multi criteria evaluation in evaluation in which knowledge of the experts is used to define the factors characterizing the watershed and the logic relations between the factors. l b h f
The knowledge base encapsulates the assessment criteria and the relationships in an explicit form so criteria and the relationships in an explicit form so that they can be easily examined, modified, or updated.
Kn ow le dge Ba se d Syst e m s
The knowledge base contains knowledge & experience for the subject domain (domain knowledge) & specifies logical relations among topics of interest to an logical relations among topics of interest to an assessment.
Inference engine performs knowledge-based
approximate reasoning to draw conclusions about the approximate reasoning to draw conclusions about the state of the system.
By integrating knowledge-based reasoning into a GIS environment provide decision support for watershed management.
GIS application provides GIS application provides database management, spatial database management spatial analysis, system interface, and map display.
Assessment system allows user to evaluate knowledge
b base for a specific spatial database & view the results. f ifi i l d b & i h l
Kn ow le dge Ba se St r u ct u r e
Knowledge base structure is a hierarchy of
dependency networks. Each network evaluates a dependency networks. Each network evaluates a specific proposition about the state of watershed condition.
Knowledge base structure Knowledge base structure is designed to address the is designed to address the
issues concerned by the watershed managers and to reflect their opinions on the importance of each issue.
At the top of the
hierarchy is the network wat ershed
for the proposition that the overall condition
condit ion
of the watershed is suitable for sustaining healthy of the watershed is suitable for sustaining healthy populations of the native
The wat ershed condit ion network depends on two o e e e e o lower level networks: st ream condit ion and upland s s ea co d o a d up a d
condit ion Multi Criteria Decision Analysis (MCDA)
Simulation Models Simulation Models of various hydrologic components of of various hydrologic components of
watershed - integrated with AI tools so as to make use of experts’ knowledge & heuristic information in decision making process; ki used to help d t h l the end users to arrive at the th d t i t th best suitable decisions related to irrigation management.
Irrigation assessment & management g g - multi-criteria
decision analysis ( MCDA ) problems - use knowledge-based systems.
MCDA MCDA - in which the land suitability criteria, water in which the land suitability criteria water
availability & irrigation requirement - various criteria to be evaluated- objective max. agricultural production.
MCDA models -used in irrigation management to identify
areas that can be irrigated, water release during different time period & best suitable cropping pattern for area p pp g p
Typical Knowledge Based Systems for WM
SCS-CN-based model Soil moisture balance Fuzzy membership approach Crop water requirement Water availability
Irrigation requirement Land suitability Water harvesting potential SMCDA for identifying the most suitable cropping zones SM CD A – Spa t io t e m por a l M u lt i Cr it e r ia D e cision An a lysis p p y Re f: Re sh m ide vi ( 2 0 0 8 ) , Ph D t h e sis, D e pt . Civil, I I T Bom ba y Fuzzy Logic Based System
Lotfi Zadeh, first presented fuzzy logic in the mid-1960's, at
- the University of California at Berke
Zadeh developed fuzzy logic as a way of processing data Zadeh developed fuzzy logic as a way of processing data later on he introduced idea of partial set membership
- Definition of fuzzy : “not clear, distinct, or precise”
- Definition of fuzzy logic
- Fuzzy Logic (FL) is a multi-valued logic that allows
intermediate values to be defined between conventional intermediate values to be defined between conventional evaluations like true/false, yes/no, high/low, etc.
Fuzzy Fuzzy Probability ≠ Probability
Probability deals with uncertainty and likelihood
- Fuzzy logic deals with ambiguity and vagueness
- Ref: L.A. Zadeh (1965) Fuzzy sets. Information and Control 8 (3) 338-353.
A A B Why Fuzzy Logic?.
C C
Based on intuition and judgment
No need for a mathematical model
Provides Provides a a smooth smooth transition transition between between members members and and nonmembers
Relatively simple, fast and adaptive
Less sensitive to system fluctuations
Because of the rule based operation,
C Can implement design objectives, difficult to express l d b d ff l mathematically, in linguistic or descriptive rules
Conventional or crisp sets are binary. Conventional or crisp sets are binary.
An element either belongs to the set or doesn't. Example- -[ True, False] OR [ 0, 1] .
Ref: D. Dubois & H. Prade (1988) Fuzzy Sets and Systems. Academic Press, New York. Fuzzy Sets
Allow elements to be partially in a set
Each element is given a degree of membership in a set
A membership function is the relationship between the A membership function is the relationship between the values of an element and its degree of membership in a set
(N = Negative, P = Positive, L = Large, M = Medium, S = Small) Ref: D. Dubois & H. Prade (1988) Fuzzy Sets and Systems. Academic Press, New York. Ref: D. Dubois & H. Prade (1988) Fuzzy Sets and Systems. Academic Press, New York.
. Membership Functions
Crisp membership functions
Crisp membership functions ( μ) are either one or zero
Example- Number greater than 10
A= { x / x > 1 0 } .
Fuzzy membership functions
Fuzzy membership functions
Membership value of not only 0 or 1
The degree of truth of a statement can g range between 0 and 1
Linguistic variables are used for fuzzy measures
Examples of fuzzy measures include close, medium, heavy, light, big, small, smart, fast, slow, hot, cold, tall and short Fuzzy measures slow, hot, cold, tall and short Ref: D. Dubois & H. Prade (1988) Fuzzy Sets and Systems. Academic Press, New York. measures Fuzzy Logic Operations
Union
Intersection
Complement
the negation of the specified membership specified membership function
Ref: D Dubois & H Prade (1988) Fuzzy Sets and Systems Academic Press New York Ref: D. Dubois & H. Prade (1988) Fuzzy Sets and Systems. Academic Press, New York.
Fuzzy Logic - Applications
Fuzzy logic concepts can be applied in:
Ride smoothness control
Braking systems; High performance drives Braking systems; High performance drives
Air-conditioning systems; Digital image processing
Washing machines Washing machines
Pattern recognition in remote sensing.
Video game artificial intelligence
Graphics controllers for automated police sketchers.
Watershed related applications: Rainfall-runoff processes.
Erosive soil measurement;hydro-ecological modeling Erosive soil measurement;hydro ecological modeling over watershed; Flood forecasting; Water quality problems; Cropping & irrigation management Fuzzy Logic – Advantages & Limitations
Adva n t a ge s
- Allows the use of vague linguistic terms in the rules
- No mathematical model
- Rule based and descriptive type
Lim it a t ion s
- Difficult to estimate membership function
- Difficult to estimate membership function
- There are many ways of interpreting fuzzy rules
- Combining the outputs of several fuzzy rules and defuzzifying the output
- Combining the outputs of several fuzzy rules and
- Input
Fuzzification
- Fuzzy base rule
- Defuzzification
Fuzzy rule Data input Data output Fu z z ifica t ion D e fu z z ifica t ion Fuzzy output
- Output
Fu z z ifica t ion Fu z z ifica t ion
Fuzzifier converts a crisp input into a fuzzy variable
It converts each piece of input data to degrees of membership
The membership function is a graphical representation
The membership function is a graphical representation of the magnitude of participation
Definition of the membership functions must
- – reflects the designer's knowledge
- – provides smooth transition between member and nonmembers of a fuzzy set nonmembers of a fuzzy set
Typical shapes of the membership function Gaussian; Trapezoidal; Triangular (commonly used) p g ( y )
Fu z z y Ba se Ru le
Fuzzy based 0.2 0.4 0.6 0.8 1 Fuzzy based decision
Eg: IF height is tall and weight is medium THEN healthy (H)
Fun f U LH SH H
Eg: IF height is tall and weight is
Rule Function: F ={H,SH,LH,U}
Include all possible fuzzy relations between inputs and outputs outputs
Rules are tabulated as fuzzy words
Rules reflect expert’s decisions
Rules are expressed in the IF-THEN format
Eg:– Healthy (H); Somewhat healthy (SH); Less Healthy (LH); Unhealthy (U) Less Healthy (LH); Unhealthy (U)
D e fu z z ifica t ion D e fu z z ifica t ion
Defuzzification converts the resulting fuzzy outputs from the fuzzy inference engine to a number from the fuzzy inference engine to a number
Converting the output fuzzy variable into a unique number f f h d
Defuzzification Methods weighted average maximum membership maximum membership average maximum membership centre of gravity
Knowledge-Based Model Development
Re sh m ide vi ( 2 0 0 8 ) , Kn ow le dge ba se d M ode l for Su pple m e n t a r y I r r iga t ion Re sh m ide vi ( 2 0 0 8 ) , “Kn ow le dge - ba se d M ode l for Su pple m e n t a r y I r r iga t ion Asse ssm e n t in Agr icu lt u r a l W a t e r sh e ds” Ph D t h e sis, D e pt . Civil, I I T Bom ba y Fuzzy rule-based inference system for land suitability evaluation suitability evaluation
SMCDA model for identifying the scope for supplementary irrigation supplementary irrigation
Graphical user interface
5 Steps in the Model Development
5 Steps in the Model Development
- – Fuzzification of the attributes
- – Estimation of the intermediate land suitability index
- – Generation of the fuzzy rule base
- – Aggregation of the rules (Fuzzy output in terms of 5 suitability classes) suitability classes)
- – Defuzzification
Problem with large number of attributes
Problem with large number of attributes
Hierarchical classification is adopted
Considered both land potential and water potential Fuzzification
Attribute values are mapped into [0,1]
Two types of attributes: Thematic attributes for land potential Unique membership value for each class; Continuously expressed attributes for land potential expressed attributes for land potential Semantic import Semantic import membership function; Asymmetric left (AL), Asymmetric right (AR) or Optimal range (OR)
I I n t e r m e dia t e La n d Su it a bilit y I n de x t di t L d S it bilit I d
Weighted aggregation of the attribute membership values
Att ib t i ht U i Attribute weights – Using Saaty’s relative importance scale S t ’ l ti i t l (Saaty, 1980)
Relative importance is assumed based on literature, field Relative importance is assumed based on literature field observation and heuristic information
Gives intermediate LSI in three suitability classes (Good, Gives intermediate LSI in three suitability classes (Good, Moderate and Not-suitable) based on land & water potential
Fuzzy Rule Base and Aggregation of the Rules
Su it a bilit y cr it e r ia
- – Ex pr e sse d in t h e for m of I F..TH EN r u le s
- – I n t e r m s of in t e r m e dia t e su it a bilit y in dice s AND good is terrain good is potential water AND good is LU IF
- )- Maximum centroid method
-
- – A change in the cropping pattern is proposed A change in the cropping pattern is proposed
- – Replaces the existing crop with the higher priority one
- – At t r ibu t e s
- – At t r ibu t e su it a bilit y for diffe r e n t cr ops
- – Cr op pr ior it y & a gr icu lt u r a l pr a ct ice s
- – La n d su it a bilit y cr it e r ia
- D r a in a ge m a p D i
- La n d u se m a p>Con t ou r m a p
- D r a in a ge de n sit y m a p
- Soil m a p
- Pr ox im it y t o w a t e r body
- pH m a p
- Pr ox im it y t o se t t le m e n t
- M a ps sh ow in g spa t ia l va r ia t ion in EC, Sa lin it y e t c.
- Ra in fa ll
- Re la t ive h u m idit y>St r e a m flow
- Su n sh in e du r a t ion>Te m pe r a t u r e
- W in d spe e d
- * % Area Suitability class Range of membership values % Area
AND moderate is terrain moderate is potential water AND good is LU IF THEN excellent is area the AND good are parameters other Good is stics characteri chemical - physico AND AND good is terrain good is potential water AND good is LU IF AND suitable is terrain suitable not is potential water AND suitable not is LU IF THEN moderate is area the AND moderate are parameters other moderate is stics characteri chemical - physico AND THEN suitable not is area the AND suitable not are parameters other suitable not is stics characteri chemical - physico AND
Generate fuzzy output in terms of 5 suitability classes (Excellent Good Moderate less suitable and Not suitable) (Excellent, Good, Moderate, less-suitable and Not-suitable)
D e fu z z ifica t ion : Convert the fuzzy output into a single value (LSI
Relative importance & LSI Relative importance & LSI
Three cases:
Case 1- LSI * of existing crop < another crop of higher priority; higher priority crop is selected g p y g p y p
Case 2- LSI * of the existing crop < another crop of lesser priority
Change in the cropping pattern if less suit able or not suit able for the existing crop
Case 3- LSI * is same for more than one potential crop
If less suit able or not suit able for the existing crop, and if relative importance of the existing crop < the other crop,
SMCDA Model for Irrigation Feasibility Analysis
Irrigation requirement Runoff availabilityRunoff Runoff
Vs
Water deficit periods
Irrigation requirement
Runoff
Vs Vs
Land suitability map L d it bilitIrrigable areas I i bl
Priority-based Irrigation requirement
Outsource water requirement
for different suitability classes for different suitability classes
Ca se St u dy: Harzul Watershed
Reshmidevi (2008), “Knowledge-based Model for Supplementary Irrigation Assessment in Agricultural
Watersheds” PhD thesis, Dept. Civil, IIT BombayLocation- Nashik district, Maharashtra, India T i l h id li t Tropical humid climate
Area: 10.9 sq. km P i i l
Paddy and finger millet
Principle crops:
Paddy and finger millet Reshmidevi, T.V., Eldh o T.I ., Jana, R., (2010 “Knowledge-Based Model for Supplementary Irrigation Assessment in Agricultural Watersheds” Jou r n a l of I r r iga t ion a n d D r a in a ge , ASCE, 2010, Vol. 136, No.6, pp. 376-382. Jou r n a l of I r r iga t ion a n d D r a in a ge , ASCE, 2010, Vol. 136, No.6, pp. 376 382. Generation of the Database Generation of the Database
H e u r ist ic in for m a t ion & fie ld obse r va t ion
M a p la ye r s
p p
P i it t t t l t
H ydr o- m e t e or ologica l da t a H ydr o- m e t e or ologica l da t a
Case St d Land S itabilit Case Study: Land Suitability Model6
(
300 100 200u
n
o
ff
mm)
Rainfall 100 200 a in fa ll ( m m ) Observed Q SimulatedQ 1006/1 7/1 8/1 9/1 Day
R
u
300 400 R Simulated Q a Ye a r 1 9 9 7 – SCS- CN - SM S- SCS- CN m ode l in cor por a t in g Soil M oist u r e Sim u la t ion 2002 3) 10 m 0.04 m 3)Model 6 ) 300 400 m 200 2002 150 m m)
Rainfall hyetograph
2002 2 4 6 8 (M il li o n m 0.01 0.02 0.03 (M il li o n m 100 2006/ 1 7/ 1 8/ 1 9/ 1 10/ M ( 1 Day m m 50 100 6/ 1 7/ 1 8/ 1 9/ 1 10/ R 1 Day a in fa ll ( m 6/ 1 7/ 1 8/ 1 9/ 1 10/ 1 Day Irrigation requirement Accumulated runoff Day
Year 2002, Dry year Paddy field Day g q ccu u ated u o
Irrigation requirement in the Harsul watershed Irrigation requirement in the Harsul watershed
2001
2002
2003 2003 2004 Be st su it a ble cr oppin g z on e s Land Suitability for Crops in Harsul Watershed La n d su it a bilit y for pa ddy La n d su it a bilit y for fin ge r m ille t Suitability class Range of LSI
La n d su it a bilit y for pa ddy La n d su it a bilit y for fin ge r m ille t Not suitable 0 - 30 85 Less suitable 30 - 45 2 Moderately suitable 45 - 60 7 Not suitable 0 - 30 Moderately suitable 45 - 60 78 Less suitable 30 - 45 1 Good 60 - 80 6 Excellent 80 - 100 Good 60 - 80 15 Excellent 80 - 100 6 Knowledge Based Modeling–Concluding Remarks
Many decision-making and problem-solving tasks are too
easy to solve in the recent days using knowledge based
system & Fuzzy Logic. t & F L i
Fuzzy logic provides an alternative way to represent
linguistic and subjective attributes of the real world in
computing computing It is able to be applied to control systems and other
applications in order to improve the efficiency and simplicity
of the design process of the design process
Design objectives difficult to express mathematically can be
incorporated in a fuzzy controller by linguistic rules. The knowledge based model shows the irrigation requirement The knowledge-based model shows the irrigation requirement
for the predicted rainfall- helps to choose / adopt appropriate
crops & irrigation management plan Re fe r e n ce s Re fe r e n ce s L.A. Zadeh (1965) Fuzzy sets. Information and Control 8 (3) 338353.
G.J. Klir, Bo Yuan (Eds.) (1996) Fuzzy Sets, Fuzzy Logic, and Fuzzy Systems. Selected Papers by Lotfi A. Zadeh. World Scientific, Singapore. Gokmen Tayfur, Serhan Ozdemir, Vijay P. Singh(2003) Advances in Water Resources G k T f S h O d i Vij P Si h(2003) Ad i W R 26 1249–1256. Chang NB, Weu GG, Chen YL.( 1997-99) A fuzzy multi-objective programming
approach for optimal management of the reservoir watershed. Eur J Oper Res (2-
1),289–302.
Yu P-S, Yang T-C.( 2000) Fuzzy multi-objective function for rainfall runoff model
calibration. J Hydrol;238(1–2), 1–14. D. Dubois and H. Prade (1988) Fuzzy Sets and Systems. Academic Press, New York. D Dubois and H Prade (1988) Fuzzy Sets and Systems Academic Press New York Reshmidevi, T.V., Eldh o T.I ., Jana, R., (2010 “Knowledge-Based Model for Supplementary Irrigation Assessment in Agricultural Watersheds” Jou r n a l of R Re sh m ide vi ( 2 0 0 8 ) , “Kn ow le dge - ba se d M ode l for Su pple m e n t a r y I r r iga t ion h id i ( 2 0 0 8 ) “K l d b d M d l f S l t I r r iga t ion a n d D r a in a ge , ASCE, 2010, Vol. 136, No.6, pp. 376-382.
I i t i
Sa a t y, T. L. ( 1 9 8 0 ) . Th e An a lyt ic H ie r a r ch y Pr oce ss, M cGr a w - H ill Pu blish in g
Asse ssm e n t in Agr icu lt u r a l W a t e r sh e ds” Ph D t h e sis, D e pt . Civil, I I T Bom ba y Com pa n y, N e w Yor k .Tu t or ia ls - Qu e st ion !.?
Critically study the applications of
knowledge based systems for various water knowledge based systems for various water
resources management problems. Study various case studies available in literature (details can be obtained from Internet). Study the role of knowledge based modeling Study the role of knowledge based modeling y y g g g g
in Integrated Water Resources Management. in Integrated Water Resources Management
Prof. T I Eldho, Department of Civil Engineering, IIT Bombay
Se lf Eva lu a t ion - Qu e st ion s!. Q
Describe the features of typical knowledge based models. models
Illustrate the requirements of knowledge based systems. systems.
Describe typical knowledge based system for watershed management. g
Illustrate the fuzzy logic operators used in typical fuzzy logic.
What are the important components of a fuzzy systems.
Prof. T I Eldho, Department of Civil Engineering, IIT Bombay
Assign m e n t - Qu e st ion s?. g Q
Describe the structure of a knowledge based system.
What are the important features of Multi Criteria Decision Analysis (MCDA).
Ill Illustrate the features of fuzzy logic based t t th f t f f l i b d
systems. Describe applications, advantages & Describe applications advantages & limitations of Fuzzy Logic?.
Illustrate a typical Knowledge based model for watershed management.
Un solve d Pr oble m !. Un solve d Pr oble m !
Critically study a typical knowledge based Critically study a typical knowledge based model for the water and land management model for the water and land management model for the water and land management model for the water and land management in a watershed. in a watershed.
For your watershed area, study the scope of For your watershed area study the scope of For your watershed area study the scope of For your watershed area, study the scope of
development of knowledge based model development of knowledge based model considering rainfall, various crops, land use, considering rainfall, various crops, land use, g g , , p , p , , , land suitability, water requirement etc. land suitability, water requirement etc.Prof. T I Eldho, Department of Civil Engineering, IIT Bombay
Dr. T. I. Eldho Dr. T. I. Eldho Professor, Professor, Department of Civil Engineering, Department of Civil Engineering, p p g g g g Indian Institute of Technology Bombay, Indian Institute of Technology Bombay, Mumbai, India, 400 076. Mumbai, India, 400 076. Email: Email: Email: Email: eldho@iitb.ac.in eldho@iitb.ac.in eldho@iitb.ac.in eldho@iitb.ac.in Phone: (022) – Phone: (022) – 25767339; Fax: 25767302 25767339; Fax: 25767302